linux-kernel - Re: [PATCH v6 3/3] clk: si5351: allow PLLs to be adjusted without reset

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <ed6a16e092feb62366d3ddbeb3cbbe64.sboyd@kernel.org>
Date: Sun, 17 Dec 2023 22:32:49 -0800
From: Stephen Boyd <sboyd@...nel.org>
To: Alvin Šipraga <alsi@...g-olufsen.dk>, Alvin Šipraga <alvin@...s.dk>, Conor Dooley <conor+dt@...nel.org>, Krzysztof Kozlowski <krzysztof.kozlowski+dt@...aro.org>, Michael Turquette <mturquette@...libre.com>, Rob Herring <robh+dt@...nel.org>
Cc: Sebastian Hesselbarth <sebastian.hesselbarth@...il.com>, Rabeeh Khoury <rabeeh@...id-run.com>, Jacob Siverskog <jacob@...nage.engineering>, Sergej Sawazki <sergej@...dac.com>, linux-clk@...r.kernel.org, devicetree@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v6 3/3] clk: si5351: allow PLLs to be adjusted without reset

Quoting Alvin Šipraga (2023-11-24 05:17:44)
> From: Alvin Šipraga <alsi@...g-olufsen.dk>
> 
> Introduce a new PLL reset mode flag which controls whether or not to
> reset a PLL after adjusting its rate. The mode can be configured through
> platform data or device tree.
> 
> Since commit 6dc669a22c77 ("clk: si5351: Add PLL soft reset"), the
> driver unconditionally resets a PLL whenever its rate is adjusted.
> The rationale was that a PLL reset was required to get three outputs
> working at the same time. Before this change, the driver never reset the
> PLLs.
> 
> Commit b26ff127c52c ("clk: si5351: Apply PLL soft reset before enabling
> the outputs") subsequently introduced an option to reset the PLL when
> enabling a clock output that sourced it. Here, the rationale was that
> this is required to get a deterministic phase relationship between
> multiple output clocks.
> 
> This clearly shows that it is useful to reset the PLLs in applications
> where multiple clock outputs are used. However, the Si5351 also allows
> for glitch-free rate adjustment of its PLLs if one avoids resetting the
> PLL. In our audio application where a single Si5351 clock output is used
> to supply a runtime adjustable bit clock, this unconditional PLL reset
> behaviour introduces unwanted glitches in the clock output.
> 
> It would appear that the problem being solved in the former commit
> may be solved by using the optional device tree property introduced in
> the latter commit, obviating the need for an unconditional PLL reset
> after rate adjustment. But it's not OK to break the default behaviour of
> the driver, and it cannot be assumed that all device trees are using the
> property introduced in the latter commit. Hence, the new behaviour is
> made opt-in.
> 
> Cc: Sebastian Hesselbarth <sebastian.hesselbarth@...il.com>
> Cc: Rabeeh Khoury <rabeeh@...id-run.com>
> Cc: Jacob Siverskog <jacob@...nage.engineering>
> Cc: Sergej Sawazki <sergej@...dac.com>
> Signed-off-by: Alvin Šipraga <alsi@...g-olufsen.dk>
> Acked-by: Sebastian Hesselbarth <sebastian.hesselbarth@...il.com>
> ---

Applied to clk-next